INDEX

Explanations

references to the term "Formula"

oai_token-act-pair · gpt-3.5-turbo

mentions of "Formula 1" and related racing events

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

camp

-0.69

haw

-0.68

ploy

-0.66

ash

-0.65

aun

-0.65

 unloaded

-0.65

achus

-0.63

chat

-0.63

scan

-0.61

 offended

-0.61

POSITIVE LOGITS

 Formula

3.96

 formula

1.67

FIA

1.49

 Motorsport

1.43

 NASCAR

1.42

 Ferrari

1.41

 formulas

1.37

 Prix

1.33

 McLaren

1.33

 motors

1.21

Activations Density 0.012%

references to the term "Formula"

mentions of "Formula 1" and related racing events

No Comments

No Known Activations